ARTICLE IN PRESS

Ecotoxicology and Environmental Safety 72 (2009) 310–315

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety

0147-65

doi:10.1

� Corr

Ecology

E-m

journal homepage: www.elsevier.com/locate/ecoenv

Comparing ecotoxicological effect concentrations of chemicals establishedin multi-species vs. single-species toxicity test systems

Frederik De Laender a,b,�, Karel A.C. De Schamphelaere a, Peter A. Vanrolleghem b,c, Colin R. Janssen a

a Laboratory of Environmental Toxicology and Aquatic Ecology, Ghent University (UGent), Jozef Plateaustraat 22, B-9000 Gent, Belgiumb Department of Applied Mathematics, Biometrics and Process Control (BIOMATH), Ghent University (UGent), Coupure Links 654, B-9000 Gent, Belgiumc modelEAU, Departement de Genie Civil, Universite Laval, Laval, QC, Canada

a r t i c l e i n f o

Article history:

Received 27 August 2007

Received in revised form

14 July 2008

Accepted 21 July 2008Available online 5 September 2008

Keywords:

Ecological risk assessment

Microcosm and mesocosm studies

Ecosystem modelling

13/$ - see front matter & 2008 Elsevier Inc. A

016/j.ecoenv.2008.07.014

esponding author at: Laboratory of Environm

, Ghent University (UGent), Jozef Plateaustraa

ail address: f.delaender@nioo.knaw.nl (F. De L

a b s t r a c t

Most ecological effect assessment methodologies use effect concentrations derived from single-species

testing (ECx,single-species-test) as the basis to estimate ‘safe’ environmental concentrations (such as

environmental quality criteria). Here, we examined to what extent such ECx,single-species-test are

representative for population-level effect concentrations in a community setting (ECx,multi-species-test).

Data from USEPA’s ECOTOX database revealed the existence of considerable scatter around the

relationship between ECx,single-species-test (endpoint: mortality) and ECx,multi-species-test (endpoint:

population abundance). However, we demonstrate that this scatter is reduced when ECx,single-species-

test and ECx,multi-species-test are determined simultaneously and by the same research group. Indeed, if

these conditions are fulfilled, the quotient of both ECx values for invertebrates approaches 1 for

chemicals that directly target invertebrates. Unfortunately, comparable data for other classes of

chemicals and/or taxonomic groups were not found. However, theoretical ecosystem model simulations,

which confirmed the results based on the above-mentioned analysis of the ECOTOX database, indicated

that for phytoplankton, EC10,single-species-test4EC10,multi-species-test, for chemicals that directly target

invertebrates. For chemicals that directly target phytoplankton, the ecosystem model simulations

suggest that ECx,single-species-test4ECx,multi-species-test for both phytoplankton and invertebrates. Hence, our

observation based on the analysis of existing experimental data that the ECx,single-species-test is similar to

the ECx,multi-species-test may be biased by the fact that only data were available for invertebrates and for

chemicals targeting invertebrates. Experimental research is required to test the predictions made by the

model simulations for phytoplankton as well as for chemicals directly targeting phytoplankton.

& 2008 Elsevier Inc. All rights reserved.

1. Introduction

Environmental programs such as Registration, Evaluation andAuthorization of Chemicals, REACH (European Commission,2004), the Canadian Environmental Protection Act, CEPA and theDomestic Substances List, DSL (Environment Canada, 2003), andthe USEPA high production chemicals assessments (Walker et al.,2004) encompass the risk assessment of large numbers ofchemicals (more than 100,000). Assessing the risk that a chemicalposes includes assessing the ecological effects of this chemical onaquatic ecosystems. One way to investigate ecological effects of achemical is to expose an artificial ecosystem to differentconcentrations of the chemical while monitoring one or morebiological parameters. Such multi-species toxicity tests, often

ll rights reserved.

ental Toxicology and Aquatic

t 22, B-9000 Gent, Belgium.

aender).

termed microcosm or mesocosm experiments, have features thatpromote their use as a realistic way of assessing chemical-causedstress (Boxall et al., 2002). The presence of multiple species aswell as time-variable exposure scenario’s and physicochemicalwater characteristics, all representative of field conditions, areexamples of such features. Unfortunately, the amount of time andresources required to conduct such experiments impedes theirroutine use (Newman and Unger, 2003). Therefore, ecologicaleffect assessments of chemicals often have to rely on ecotoxicitytests with single species, despite their lower degree of realism(Sanderson et al., 2004; Schindler, 1998). Usually, in mostecological effect assessment techniques, a set of single-speciestoxicity test results (ECx,single species) is then extrapolated to thecommunity level to estimate a ‘safe’ environmental concentrationof the chemical for an ecosystem or the potentially affectedfraction of species at a given concentration of the chemical(Aldenberg and Slob, 1993; Vanstraalen and Denneman, 1989;Wagner and Lokke, 1991). Obviously, the accuracy of theassessment with respect to effects occurring in the field will

ARTICLE IN PRESS

1e−02

ECx, multi−species (µg/l)

EC

x, sin

gle

−specie

s (

µg/l)

NOECsLOECsEC50s

1e−02 1e+021e+00

1e+00

1e+02

Fig. 1. A comparison between ECx,multi-species-test and ECx,single-species-test using data

from the aquatic part of the USEPA database ECOTOX. Only data were included for

which species scientific name and effect magnitude were identical for both ECx

values. Different symbols denote different types of effect concentrations, as given

by the legend.

F. De Laender et al. / Ecotoxicology and Environmental Safety 72 (2009) 310–315 311

increase if the ECx,single-species-test values used as the basis, of suchextrapolations are similar to effect concentrations observed in amulti-species field situation (ECx,multi-species), i.e. if ECx,single-species-

test ¼ ECx,multi-species. There are, however, a number of biologicaland physicochemical reasons as to why this may not be true.

In a single-species toxicity test, the direct effect of a chemicalon individuals of a single species is usually evaluated undercontrolled laboratory conditions. The resulting effect concentra-tion, i.e. an ECx,single-species-test, reflects the susceptibility of themonitored individuals to direct chemical effects. In multi-speciestoxicity tests such as microcosm and mesocosm experiments,the net result of direct and indirect effects is evaluated on apopulation level. The term ‘indirect effect’ is used here as inFleeger et al. (2003). Examples of indirect effects include, but arenot limited to, (i) an increased abundance of prey resulting from adecreased abundance of a predator due to a direct effect ofa chemical on this predator, and (ii) a decreased abundance of apredator resulting from a decreased abundance of prey caused bya direct toxic effect on this prey. These examples clearly show thatindirect effects may be both (numerically) positive (increasedabundance) as well as negative (decreased abundance). Apartfrom the realm of predator–prey relationships in multi-speciesexperiments, also differences in physicochemical conditions of theexposure media may cause ECx,single-species-test to be different fromECx,multi-species-test. For example, physicochemical properties likethe dissolved organic carbon concentration (De Schamphelaereand Janssen, 2004) and temperature (Heugens et al., 2001) havebeen shown to considerably alter the magnitude of direct effectsof chemicals.

With the present study, we, wished to investigate, using adatabase-analysis approach, the implicit assumption of manyecological effects assessments that ECx,multi-species-test ¼ ECx,single-

species-test. We did this by searching the USEPA database ECOTOXfor population-level ECx,multi-species-test values and correspondingECx,single-species-test values for the same chemical and species.

2. Material and methods

The search for ECx,multi-species-test values in the aquatic part of the USEPA

database ECOTOX (http://cfpub.epa.gov/ecotox/) was constrained by selecting

concentration-based endpoints describing effects on a population level in fresh-

water (semi-)field tests. All data between 1915 and 2007 were considered. Both

animal and plant data were collected. Population abundance and biomass

constituted up to 88% of all reported endpoints within the population group.

Corresponding ECx,single-species-test values were sought by constraining our

ECOTOX search to concentration-based endpoints describing effects in freshwater

single-species tests. CAS numbers of the chemicals for which ECx,multi-species-test

values were found were given as chemical entry to ensure correspondence. Again,

all data between 1915 and 2007 were considered. Both animal and plant data were

collected. Studies on benthic communities were omitted. The majority of the

single-species toxicity data (85%) described the effects on survival of invertebrates.

Results of both described search operations were merged and a filtering

operation was performed to ensure that species scientific names and the effect-

magnitude (x in ECx) were identical for ECx,multi-species-test and ECx,single-species-test. In

cases where a NOEC or a LOEC instead of an ECx was reported, the effect-

magnitude can be considered, literally, ‘‘no observable effect’’ or ‘‘lowest

observable effect’’, respectively, thus ensuring that NOEC and LOEC values were

also useful for our analysis. In this way, we obtained a dataset where every record

consisted of (1) a species’ scientific name, (2) its ECx,single-species-test and (3) its

ECx,multi-species-test. All operations were performed using the free software R (http://

www.r-project.org/).

3. Results

The dataset we obtained consisted of 49 records of whichseven were insects and spiders and 42 were crustaceans (25represented Daphnia pulex). Most ECx,multi-species-test were within afactor two of the corresponding ECx,single-species-test (Fig. 1). One

NOEC and two LOECs were clearly located below the line oftwofold error (up to two orders of magnitude difference),indicating that N/LOEC,multi-species-test4N/LOEC,single-species-test.Similarly, one NOEC and one LOEC are clearly located abovethe factor two-line and hence indicate N/LOEC,multi-species-test

oN/LOEC,single-species-test. However, note that for the datasetpresented in Fig. 1, confidence intervals were not availablewhich makes it difficult to be decisive on whether or notECx,multi-species-test significantly differed from ECx,single-species-test.

The dataset of 49 records only contained ECx data for 2chemicals (‘‘1H-benzimidazol-2-yl carbamic acid, methyl ester’’,and ‘‘phosphorothioic acid, O,O-diethyl O-(3,5,6-trichloro-2-pyridinyl) ester’’), which is too limited to meaningfully addressthe questions posed here. The most limiting constraint posedon the ECOTOX search results appeared to be the requirementthat reported species scientific name for ECx,multi-species-test andECx,single-species-test should be identical. Following closer inspectionof the search results, we found that scientific names in ECOTOXwere not always specified at the species level. This resulted in,for example, Daphnia sp. to be judged as taxonomically differentfrom Dapnia magna by the filtering operations. Hence, the limitedoutput of the filtering operations was primarily caused bydifferent levels of the reported taxonomical resolution. Tocircumvent this problem, we redefined the filtering constrainton species equality for ECx,multi-species-test and ECx,single-species-test.Instead of using the attribute ‘species scientific name’, we usedthe attribute ‘species common name’ to decide whether the samespecies were used for ECx,multi-species-test and ECx,single-species-test

derivation. This yielded 610 extra records and 3 more chemicalsfor which ECx,multi-species-test and ECx,single-species-test could becompared: ‘‘(1 alpha, 2 alpha, 3 beta, 4 alpha, 5 alpha, 6 beta)-1,2, 3, 4, 5, 6-hexachlorocyclohexane’’, ‘‘2, 3 ,4, 6-tetrachlorophe-nol’’, and ‘‘4-nonylphenol’’. The majority of the examined specieswere crustaceae (83%), other species were insects and amphipods.Differences between EC50,multi-species-test and EC50,single-species-test

were in most cases limited to a factor 2 (Fig. 2), while differencesbetween N/LOECmulti-species-test and N/LOECsingle-species-test showmore deviation from the 1:1 line. We found that N/LOECmulti-species

was higher than N/LOECsingle-species in 90% of the cases (Fig. 3).

ARTICLE IN PRESS

1e−02

ECx, multi−species (µg/l)

EC

x, sin

gle

−specie

s (

µg/l)

NOECs

LOECs

EC50s

1e−02 1e+00 1e+02

1e+00

1e+02

Fig. 2. A comparison between ECx,multi-species-test and ECx,single-species-test using data

from the aquatic part of the USEPA database ECOTOX. Only data were included for

which species common name and effect magnitude were identical for both

ECx values. Different symbols denote different types of effect concentrations, as

given by the legend.

xx xx xxxx xxxxxxxxxx x xxxxxxx

xx xxxxxxxxxxxxxxxxxxxxxxxx xx xxx xxxxx xxxxxxx

xxxxxxxxxxxx xxxxxxx

xxxxxxxx xxxxx xxxxx xxxxxxx

xxxxxxxxxxxxxxxxxxx xxxxx xxxxxxx

xxxxxxxxx xxxxxxx

xxxxxxxxxxxxxxxxxxxxxxxxxx xxx xxx xxxxxxx

xxxxxxxxxxx xxx xxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx xxxxxxx

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxx

xxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxx

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

1e−01

0.0

0.2

0.4

0.6

0.8

1.0

NOEC multi−species / NOEC single−species (−)

pro

ba

bili

ty (

−)

1e+00 1e+01 1e+02 1e+03

Fig. 3. Cumulative probability distribution of the quotient ECx,multi-species-test/

ECx,single-species-test with ‘NOEC’ as effect magnitude. Data originate from the aquatic

part of the USEPA database ECOTOX and have been filtered using the criteria given

in the caption of Fig. 2 and in the material and method section. The vertical line is

at quotient equal to 1.

1e−01

0.0

0.2

0.4

0.6

0.8

1.0

exposure time multi−species exp. /

exposure time single−species exp. (−)

pro

babili

ty (

−)

NOECsLOECsEC50s

1e+00 1e+01 1e+02 1e+03

Fig. 4. Cumulative probability distribution of the quotient of duration of multi-

species tests/duration of single-species tests. Different symbols denote different

types of effect concentrations, as given by the legend. Data originate from the

aquatic part of the USEPA database ECOTOX and have been filtered using

the criteria given in the caption of Fig. 2 and in the material and method section.

The vertical line is at the quotient equal to 1.

F. De Laender et al. / Ecotoxicology and Environmental Safety 72 (2009) 310–315312

Apparently, the (threshold) concentration resulting in (no) effectsin a multi-species experiment is usually higher than in a single-species experiment. This finding is somehow counter-intuitive asexposure duration in the considered multi-species experiments ishigher than in the single species, most notably for NOECs (Fig. 4).

4. Discussion

A comparison between ECx,multi-species-test and ECx,single-species-

test using publicly available datasets may increase our insight into

the influence that a species’ biological and physicochemicalenvironment has on the (adverse) effects it experiences whenexposed to a chemical. Unfortunately, the relationship betweenECx,multi-species-test and ECx,single-species-test exhibited a considerableamount of scatter (Fig. 2). Here,below, we will address the factorsthat may have lead to this scatter.

The magnitude of the direct effect of a chemical on a specieshas been shown not to be a constant, but variable depending on,among other things, the variability of a test population’s fitnessover time (Degraeve et al., 1991) or between genotype variation ofsensitivity (Baird et al., 1991; Barata et al., 2002). As publiclyavailable datasets include results from tests conducted bydifferent research groups, which have possibly worked withpopulations or clones with different sensitivity or by the sameresearch group, but where single-species and multi-species testshave not been conducted simultaneously, the direct effectreported to be experienced by a given species may differ betweenthe single-species and multi-species study result. Hence, it isunsure if ECx,multi-species-test and ECx,single-species-test in our previousanalysis differ, because direct effects are different between single-species and multi-species tests, or because of the presence ofindirect effects in the multi-species test. As we wanted toinvestigate reliably the indirect effects of a chemical on a speciesby comparing ECx,multi-species-test with ECx,single-species-test, the datashould come from experiments where the species experience thesame magnitude of direct effects in both experimental settings.

Therefore, in an additional analysis, we used the ISI Webof Science [TS ¼ (mesocosm* OR microcosm* OR enclosure* ORsemi-field OR semi field) AND (ecosystem* OR communit*) ANDtox* NOT soil AND effect*] to analyse primary studies in whichsingle-species toxicity tests were performed simultaneously withmulti-species experiments by a single research group. We couldonly identify 5 such studies, i.e. Fairchild et al. (1992), Hose et al.(2003), Schroer et al. (2004), van der Hoeven and Gerritsen (1997),and van Wijngaarden et al. (1996), yet comprising 32 data points.In only one study a fish species was investigated, while all otherstudies investigated invertebrates. The four chemicals involved in

ARTICLE IN PRESS

F. De Laender et al. / Ecotoxicology and Environmental Safety 72 (2009) 310–315 313

these studies, i.e. esfenvalerate, chlorpyrifos (2 studies), lambda-cyhalothrin, and endosulfan primarily target invertebrates.

The data were clearly less scattered than what was found fromthe full ECOTOX database analysis (Figs. 1 and 2). This isillustrated by the high correlation coefficient between medianECx,multi-species-tests and ECx,single-species-test (0.94), compared to thecorrelation coefficient when using the ECOTOX data (0.44). Thissuggests that temporal and genotypic (or between population)variability of sensitivity can have an influence on how therelationship between ECx,multi-species-test and ECx,single-species-test isinterpreted.

Confidence intervals of all data indicate less than a factor twodeviation from the 1:1 line, suggesting that ECx,multi-species-testsEECx,single-species-test (Fig. 5). In 3 out of the 5 studies found in theopen literature data, single-species testing was performed in cagesplaced within the multi-species enclosure (Hose et al., 2003;Schroer et al., 2004; van Wijngaarden et al., 1996). The use of cagesprevents contact with the other species in the experimentalenclosures while the same physicochemical environment in multi-and single-species tests is guaranteed. Hence, the absence of markeddifferences between ECx,single-species-test and ECx,multi-species-test inthe 5 reviewed studies suggests a limited influence of thebiological environment, i.e. predator–prey interactions, on theresponse of the organisms to a chemical in a multi-species setting.

One should, however, bear in mind the limited numberof chemicals for which this exercise was made. We found datafor only 5 and 4 chemicals with the ECOTOX and the posterior ISIweb of science approach, respectively. To examine if the resultsobtained can be generalised across chemicals, we performed atheoretical exercise with an ecosystem model methodologicallyidentical to a previous study (De Laender et al., 2008). Inthat study, we randomly assigned mortality-ECx,single-species-test

values to species from a virtual multi-species experimentconsisting of fish, zooplankton, and phytoplankton. With avalidated ecosystem model these ECx,single-species-test values weretransformed into abundance-ECx,multi-species-test values. Details onthe methodology are described in De Laender et al. (2008). In thepresent paper, we choose x ¼ 10 which allowed us to examine if

1e−04

1e−04

ECx, multi−species (µg/l)

EC

x, sin

gle

−specie

s (

µg/l)

LC50s

EC50sEC10s

1e+02

1e−02

1e+00

1e+02

1e−02 1e+00

Fig. 5. A comparison between ECx,multi-species-test and ECx,single-species-test using data

from an open literature search. Again, only data were included for which species

common name and effect magnitude were identical for both ECx values. Care was

taken that single-species and multi-species tests were performed by the same

authors at the same time. Different symbols denote different types of effect

concentrations, as given by the legend.

EC10,single-species-test ¼ EC10,multi-species-test. The advantage of such anapproach is that the EC10,single-species-test values of the species inthe virtual multi-species experiment are a priori known as theyare parameters of the ecosystem model. Hence, we were surethat species in the (virtual) multi-species test experiencedthe same direct effects as in the (virtual) single-species test.Hence, a deviation of EC10,multi-species-test from EC10,multi-species-test

indicated the occurrence of indirect effects. Additionally, theuse of a modelling approach allows to check if EC10,single-species-test

¼ EC10,multi-species-test for many different chemicals. In line withthe finding from our literature search (Fig. 5), the eco-system model simulations suggested that EC10,multi-species-tes andEC10,single-species-test for invertebrates did not differ more thana factor 2 (Fig. 6) and did not exhibit a lot of scatter. Note that inFig. 6, only simulations for chemicals that primarily targetinvertebrates have been compiled, i.e. those chemicals for whichlog(EC10,single-species-test, invertebrates)�log(EC10,single-species-test, phytoplankton)p�1. These chemicals are representative for those used in ourdatabase-based analysis described in ‘Results’.

The majority of toxicity test results, whether they are derivedin a single- or multi-species test, use invertebrates as a studyobject and mortality as an endpoint (Breitholtz et al., 2006).Hence, it is not surprising that our ECOTOX database approach aswell as our literature search only reflects data on animals, themajority of which were invertebrates. However, phytoplanktonblooms as an indirect effect from exposure to insecticides areoften observed in experimental ecosystems (Hanazato, 2001).When EC10,multi-species-test and EC10,single-species-test from our modelsimulations are plotted for chemicals targeting invertebrates (i.e.as in this study), a difference can be seen between predictionsfor phytoplankton vs. invertebrates (Fig. 6). The trend shownin Fig. 6 for invertebrates agrees well with what was found in ourpresented database-based approach. However, predictions forphytoplankton suggest that ECx,multi-species-testoECx,single-species-test.Hence, our database-based finding that ECx,multi-species-testEECx,single-species-test (Fig. 5) may be skewed, because data avail-ability itself is skewed towards invertebrates. More research isrequired, however, to investigate this statement.

1e−04

1e−04

ECx, multi−species (µg/l)

EC

x,

sin

gle

−sp

ecie

s (

µg/l)

phytoplanktoninvertebrates

1e+02

1e−02

1e+00

1e+02

1e−02 1e+00

Fig. 6. A comparison between ECx,multi-species-test and ECx,single-species-test for

chemicals that directly target invertebrates, predicted by an ecosystem model, as

in a previous exercise (De Laender et al., 2008). Different symbols denote different

taxonomic groups, as given by the legend.

ARTICLE IN PRESS

1e−04

1e−04

ECx, multi−species (µg/l)

EC

x, sin

gle

−specie

s (

µg/l)

phytoplankton

invertebrates

1e−02

1e+00

1e+02

1e−02 1e+00 1e+02

Fig. 7. A comparison between ECx,multi-species-test and ECx,single-species-test for

chemicals that directly target phytoplankton, predicted by an ecosystem model

as in a previous exercise (De Laender et al., 2008). Different symbols denote

different taxonomic groups, as given by the legend.

F. De Laender et al. / Ecotoxicology and Environmental Safety 72 (2009) 310–315314

All studies, which were re-analyzed, examined effects ofsubstances which primarily target invertebrates. For the chemicalsexamined, no phytoplankton-ECx,single-species-test values were foundthat were in the range of the exposure concentrations. In thecase of endosulfan, a 96h-EC50 of 427.8mg L�1 for photosynthesisinhibition of the phytoplankton species Pseudokirchneriella sub-

capitata was reported (De Lorenzo et al., 2002). This EC50 is wellabove the range of the endosulfan concentrations tested in theecosystem study conducted by Hose et al. (2003) (o100mg L�1).Hence, it is unlikely that phytoplankton species experienced directeffects in the latter study. Decreasing zooplankton concentrations,as a result from direct chemical effects on phytoplankton, havebeen observed, for example, for atrazine (Denoyelles et al., 1982).This suggests that for chemicals directly affecting phytoplankton,ECx,multi-species-test of invertebrates is expected to be lower than theECx,single-species-test. This is confirmed by the theoretical ecosystemmodel simulations (Fig. 7). For such chemicals, a considerablefraction of the simulated cases exhibited a ECx,multi-species-

testoECx,single-species-test, both for invertebrates as for phytoplank-ton (Fig. 7). Again, the finding from our database-based approach,i.e. that ECx,multi-species-test do not differ more than a factor twofrom ECx,single-species-test, is thus partly due to a disproportionateattention in literature for effects of chemicals that directly targetinvertebrates.

5. Conclusions

From our exercise based on the ECOTOX database, it appears asif ECx,multi-species-test and ECx,single-species-test for invertebrates do notdiffer more than a factor two for the chemicals examined, exceptfor NOECs which seem to be lower in the single-species tests. Therelationship between both ECx values was less scattered whensingle-species and multi-species tests were conducted simulta-neously and by the same research group. The relationshipbetween ECx,multi-species-test and ECx,single-species-test approachedthe 1:1 line, i.e. differences between most ECx,multi-species-test andECx,single-species-test values were limited. Because of the imposedconstraints on our compiled data, data points were only available

for 4 and 5 chemicals, using the ECOTOX database and a literaturesearch as primary data source, respectively. Theoretical ecosystemmodel simulations with different hypothetical chemicals con-firmed our present findings for chemicals that primarily targetinvertebrates.

The majority of the studies involved in our databased approachpresented here focused on invertebrates. Hence, the conclusionsdrawn should be interpreted as such, since predictions from theo-retical simulations suggest EC10,multi-species-testoEC10,single-species-test

for phytoplankton.The chemicals that were involved in our database-based

approach presented here exerted direct effects on invertebrates,while direct effects on phytoplankton most likely did not occur.Theoretical ecosystem model simulations suggested EC10,multi-

species-testoEC10,single-species-test, for all populations involved whenexposed to chemicals that directly affect phytoplankton. Theseresults illustrate that our interpretation of the relationshipbetween ECx,multi-species-test and ECx,single-species-test may have beenlargely determined by data availability.

Acknowledgments

Frederik De Laender is a recipient of a Ph.D. grant of the FlemishInstitute for the Promotion of Scientific and Technological Researchin Industry (IWT, Belgium). Karel De Schamphelaere is currentlysupported by a post-doctoral fellowship from the Fund for scientificResearch in Flanders (FWO-Vlaanderen, Belgium). Peter Vanrolle-ghem is Canada Research Chair in Water Quality Modelling.

References

Aldenberg, T., Slob, W., 1993. Confidence-limits for hazardous concentrations basedon logistically distributed NOEC toxicity data. Ecotoxicology Environ. Saf. 25,48–63.

Baird, D.J., Barber, I., Bradley, M., Soares, A., Calow, P., 1991. A comparative study ofgenotype sensitivity to acute toxic stress using clones of daphnia magna straus.Ecotoxicology Environ. Saf. 21, 257–265.

Barata, C., Baird, D.J., Soares, A., 2002. Determining genetic variability in thedistribution of sensitivities to toxic stress among and within field populationsof daphnia magna. Environ. Sci. Technol. 36, 3045–3049.

Boxall, A.B.A., Brown, C.D., Barrett, K.L., 2002. Higher-tier laboratory methodsfor assessing the aquatic toxicity of pesticides. Pest Manage. Sci. 58,637–648.

Breitholtz, M., Ruden, C., Hansson, S.O., Bengtsson, B.E., 2006. Ten challenges forimproved ecotoxicological testing in environmental risk assessment. Ecotox-icol. Environ. Saf. 63, 324–335.

De Laender, F., De Schamphelaere, K.A.C., Vanrolleghem, P.A., Janssen, C.R., 2008.Do we have to incorporate ecological interactions in the sensitivity assessmentof ecosystems? An examination of a theoretical assumption underlying speciessensitivity distribution models. Environ. Int. 34, 390–396.

De Lorenzo, M.E., Taylor, L.A., Lund, S.A., Pennington, P.L., Strozier, E.D., Fulton,M.H., 2002. Toxicity and bioconcentration potential of the agriculturalpesticide endosulfan in phytoplankton and zooplankton. Arch. Environ.Contam. Toxicol. 42, 173–181.

De Schamphelaere, K.A.C., Janssen, C.R., 2004. Development and field validation ofa biotic ligand model predicting chronic copper toxicity to daphnia magna.Environ. Toxicol. Chem. 23, 1365–1375.

Degraeve, G.M., Cooney, J.D., McIntyre, D.O., Pollock, T.L., Reichenbach, N.G., Dean,J.H., Marcus, M.D., 1991. Variability in the performance of the 7-day fatheadminnow (pimephales-promelas) larval survival and growth test—an intrala-boratory and interlaboratory study. Environ. Toxicol. Chem. 10, 1189–1203.

Denoyelles, F., Kettle, W.D., Sinn, D.E., 1982. The responses of plankton commu-nities in experimental ponds to atrazine, the most heavily used pesticide in theUnited States. Ecology 63, 1285–1293.

Environment Canada, 2003. Existing Substances Evaluation Bulletin. Ottawa ON.European Commission, 2004. Why do we need REACH? REACH in brief. Brussels.Fairchild, J.F., Lapoint, T.W., Zajicek, J.L., Nelson, M.K., Dwyer, F.J., Lovely, P.A., 1992.

Population-level, community-level and ecosystem-level responses of aquaticmesocosms to pulsed doses of a pyrethroid insecticide. Environ. Toxicol. Chem.11, 115–129.

Fleeger, J.W., Carman, K.R., Nisbet, R.M., 2003. Indirect effects of contaminants inaquatic ecosystems. Sci. Total Environ. 317, 207–233.

Hanazato, T., 2001. Pesticide effects on freshwater zooplankton: an ecologicalperspective. Environ. Pollut. 112, 1–10.

ARTICLE IN PRESS

F. De Laender et al. / Ecotoxicology and Environmental Safety 72 (2009) 310–315 315

Heugens, E.H.W., Hendriks, A.J., Dekker, T., van Straalen, N.M., Admiraal, W., 2001.A review of the effects of multiple stressors on aquatic organisms and analysisof uncertainty factors for use in risk assessment. Crit. Rev. Toxicol. 31, 247–284.

Hose, G.C., Hyne, R.V., Lim, R.P., 2003. Toxicity of endosulfan to Atalophlebia spp.(ephemeroptera) in the laboratory, mesocosm, and field. Environ. Toxicol.Chem. 22, 3062–3068.

Newman, M.C., Unger, M.A., 2003. Fundamentals of Ecotoxicology, second ed.Lewis Publisher, Boca Raton.

Sanderson, H., Boudreau, T.M., Mabury, S.A., Solomon, K.R., 2004. Effects ofperfluorooctane sulfonate and perfluorooctanoic acid on the zooplanktoniccommunity. Ecotoxicol. Environ. Saf. 58, 68–76.

Schindler, D.W., 1998. Replication versus realism: the need for ecosystem-scaleexperiments. Ecosystems 1, 323–334.

Schroer, A.F.W., Belgers, J.D.M., Brock, T.C.M., Matser, A.M., Maund, S.J., Van denBrink, P.J., 2004. Comparison of laboratory single species and field population-level effects of the pyrethroid insecticide lambda-cyhalothrin on freshwaterinvertebrates. Arch. Environ. Contam. Toxicol. 46, 324–335.

van der Hoeven, N., Gerritsen, A.A.M., 1997. Effects of chlorpyrifos on individualsand populations of daphnia pulex in the laboratory and field. Environ. Toxicol.Chem. 16, 2438–2447.

van Wijngaarden, R.P.A., van den Brink, P.J., Crum, S.J.H., Voshaar, J.H.O., Brock,T.C.M., Leeuwangh, P., 1996. Effects of the insecticide Dursban(R) 4E (activeingredient chlorpyrifos) in outdoor experimental ditches.1. Comparison ofshort-term toxicity between the laboratory and the field. Environ. Toxicol.Chem. 15, 1133–1142.

Vanstraalen, N.M., Denneman, C.A.J., 1989. Ecotoxicological evaluation of soilquality criteria. Ecotoxicol. Environ. Saf. 18, 241–251.

Wagner, C., Lokke, H., 1991. Estimation of ecotoxicological protection levels fromNOEC toxicity data. Water Res. 25, 1237–1242.

Walker, J.D., Knaebel, D., Mayo, K., Tunkel, J., Gray, D.A., 2004. Use of QSARs topromote more cost-effective use of chemical monitoring resources. 1. Screen-ing industrial chemicals and pesticides, direct food additives, indirect foodadditives and pharmaceuticals for biodegradation, bioconcentration andaquatic toxicity potential. Water Qual. Res. J. Can. 39, 35–39.